Using Psychological Characteristics of Situations for Social Situation Comprehension in Support Agents

Abstract

Support agents that help users in their daily lives need to take into account not only the user's characteristics, but also the social situation of the user. Existing work on including social context uses some type of situation cue as an input to information processing techniques in order to assess the expected behavior of the user. However, research shows that it is important to also determine the meaning of a situation, a step which we refer to as social situation comprehension. We propose using psychological characteristics of situations, which have been proposed in social science for ascribing meaning to situations, as the basis for social situation comprehension. Using data from user studies, we evaluate this proposal from two perspectives. First, from a technical perspective, we show that psychological characteristics of situations can be used as input to predict the priority of social situations, and that psychological characteristics of situations can be predicted from the features of a social situation. Second, we investigate the role of the comprehension step in human-machine meaning making. We show that psychological characteristics can be successfully used as a basis for explanations given to users about the decisions of an agenda management personal assistant agent.

Who’s That? - Social Situation Awareness for Behaviour Support Agents

“…The studies described in the previous section provide evidence indicating that the proposed three-level social situation awareness component is implementable in practice [Kola et al, 2021[Kola et al, , 2020b. Their results are based on user studies revolving around the use case of a socially aware personal assistant agent for calendar management.…”

“…An advantage of a multi-level architecture is that explanations can be given on different levels: the agent can (1) give insight on the suggestion relating it to a certain personal value or preferred behavior (Level 3), (2) explain why a certain behavior or personal value is expected in a specific social situation by referring to the psychological characteristics of that situation (Level 2), and (3) give further insight on the situation cues and social relationship aspects that cause the situation to have those specific psychological characteristics (Level 1). Kola et al [2021] show through a user study that people find explanations based on Level 1 and Level 2 information to be satisfying and complete.…”

“…Human-grounded evaluations involve real people who are presented with simplified tasks, and are particularly useful in cases such as ours where the goal is to evaluate reasoning components and a full fledged agent cannot yet be implemented due to open challenges in interaction modules. In Kola et al [2021], we showed that transitioning through the three levels of the architecture is possible: using data collected from a large user study, we presented an approach in which it is possible to predict Level 2 information from Level 1 inputs, and then in turn use the predicted Level 2 information as input for predicting Level 3 information. Furthermore, we showed how Level 1 and Level 2 information can be used as a basis for creating explanations that are satisfying for people.…”

“…A technical requirement of the architecture is the ability to derive these psychological characteristics from the information from Level 1. This was investigated in Kola et al [2021]. In that work we collected Level 1 and 2 data through a crowdsourcing user study.…”

“…Further, this shows that transition from Level 2 to 3 is feasible with respect to personal values. To show that this transition is also possible in terms of expected behaviors, in Kola et al [2021] we used psychological characteristics of situations as input to predict expected user behavior regarding social priorities with an error of 0.98 on a 7-point Likert scale for actual values of the characteristics, and with an error of 1.37 for predicted values of psychological characteristics based on Level 1 information.…”